In a database management system (DBMS), query estimators use variations of probabilistic counting to determine a query path. While these estimators take into account skewness and other lower order moments of the sampled data, they ignore other characteristics of data distribution and, as such, can produce inaccurate and inefficient query plans.